• glibc 2.41 testsuite failure on powerpc

    From Aurelien Jarno@21:1/5 to All on Wed Feb 5 23:10:01 2025
    XPost: linux.debian.ports.powerpc

    Dear powerpc porters,

    The glibc 2.41-1 (version in experimental) testsuite fails to pass on
    powerpc:

    | FAIL: math/test-float-log10
    | original exit status 1
    | testing float (without inline functions)
    | Failure: Test: log10_downward (0x1p+0)
    | Result:
    | is: -0.00000000e+00 -0x0.000000p+0
    | should be: 0.00000000e+00 0x0.000000p+0
    | difference: 0.00000000e+00 0x0.000000p+0
    | ulp : 0.0000
    | max.ulp : 3.0000
    |
    | Test suite completed:
    | 4 max error test cases,
    | 148 input tests,
    | - with 600 tests for exception flags,
    | - with 140 tests for errno executed.
    | 1 errors occurred.
    | ----------
    | ----------
    | FAIL: math/test-float32-log10
    | original exit status 1
    | testing _Float32 (without inline functions)
    | Failure: Test: log10_downward (0x1p+0)
    | Result:
    | is: -0.00000000e+00 -0x0.000000p+0
    | should be: 0.00000000e+00 0x0.000000p+0
    | difference: 0.00000000e+00 0x0.000000p+0
    | ulp : 0.0000
    | max.ulp : 3.0000
    |
    | Test suite completed:
    | 4 max error test cases,
    | 148 input tests,
    | - with 600 tests for exception flags,
    | - with 140 tests for errno executed.
    | 1 errors occurred.

    The full build log is available there: https://buildd.debian.org/status/fetch.php?pkg=glibc&arch=powerpc&ver=2.41-1&stamp=1738281295&raw=0

    I have tracked down the issue to this GCC bug: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=67771

    The issue is specific to powerpc and doesn't affect ppc64 or ppc64el as
    they have a baseline that includes the fcfid instruction. Could you
    please have a look?

    Thanks
    Aurelien

    --
    Aurelien Jarno GPG: 4096R/1DDD8C9B [email protected] http://aurel32.net

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)